Applying Application Load Testing Skills to my Metabolic System

I have been amazed by the changes in my body over the past five months, as I took some practices from my Load Testing based approach for Root Cause Analysis of System problems in IT – and applied them to my own body.

In summary, I have LOST 20 Kg of weight, STOPPED snoring, have more energy and have a much better working knoweldge of what helps/hurts my body.

Information System:

If an organization engages me to perform root cause analysis serious performance and stability problems in a critical IT application then I tend to jump right in to discover as much as I can, as fast as I can, about the actual behaviour of that system. I would switch from an ‘helicopter’ to a microscopic view, as required, to get a handle on the important drivers of key metrics that may be related to the problems being experienced. I would combine passive observation with various targeted tests in a methodical, science based approach. However, such an approach often identifies issues that are completely unexpected, and the resolution of those particular problem can yield benefits much greater than anyone could reasonably expect.

Metabolic System:

 Five months back, my doctor explained that my Cholesterol results were poor, and the outlook was not good if I did not do something about it. Without change, I would probably move closer to diabetes, fatty liver disease and others I can’t (or don’t want to) remember. He suggested reducing carbohydrates, especially for the evening meal. So with my usual enthusiasm, I embraced this advice, and went on an intensive learning journey. After learning about Low Carb / High Fat diets, I decided to give it a go, and started measuring blood glucose and ketone levels frequently, as a guide to help me stay on or at least close to me desired new path. I then stepped up my learning to include more details about how we metabolise food, and how energy is stored, transported and delivered around our bodies. I started the relatively steep learning curve relating to Macronutrients, Glucose, Ketones, Triglycerides, Cholesterol and Hormones – which was, it seems, an unconscious goal that was formulated during that doctors consultation.

 Testing:

If I decided to look at how busy the CPU of a server was, and then draw conclusions from that observation – I would not likely stay in business for long, because CPU utilization fluctuates wildly, from second to second, minute to minute, hour to hour…. However, I was told that I could not have a follow up Cholesterol test for six months, as it changes very slowly, costs money and I several months need to pass before positive changes in my diet/lifestyle would be reflected in the results. However, it appeared to me, from my new found learnings, that our bodies responds dramatically to various inputs, so I obtained a CardioChek PA device along with a number of Lipid Panel strips. I have had the device for just over 2 weeks, and in that time have run more than 70 Lipid Panel tests (to show Triglycerides, Cholesterol, HDL Cholesterol and calculated LDL Cholesterol). This allowed me to use a key tool I borrowed by my Day Job, which involves drilling down to a very low granularity of measurement. This meant stepping up from 2 tests a year to 4.5 tests per day – allowing for approximately 1,000 times more granular results. I found it most interesting to show the results of the testing in terms of the time since last eating, and I combined that with Ketone Levels at the time.

 Variation:

When having my Cholesterol Test, I was asked to fast. I fasted 12 hours, but it seems that guidance varies from 8-15 hours. For that particular window, in the tests I executed, I could ‘cherry pick’ wonderful or terrible results. This was most disconcerting, from a professional perspective, as it seems that the potential treatment plan that could be recommended to me may be based on the ‘luck of the draw’ of the instantaneous measurement at the time of my blood test.

 Methodology:

This testing experiment was all about discovery and learning, so I did not attempt to ‘control’ for multiple variables. However, in the month before testing, I had established myself into a constant nutritional ketosis state by adhering to a Low Carbohydrate High Fat diet. An interesting side effect of this style of eating is that one is not hungry for snacks, so by the time I started measuring LIPIDs, I had virtually eliminated snacks, and was frequently skipping one or two meals a day, which allowed me to get a decent distribution of data-points. Some meals were extremely high in fat, and others had a moderate proportion of protein. Overall, my vegetable intake has dramatically increased for cauliflower, broccoli, eggplant, spinach and capsicum at the expense of potatoes. I think my daily carbohydrates was less than 30 grams on the majority of the days where I tested my Lipids, but the days where carbs increased above this level drove my Ketone level down, and allowed me to acquire several data points with lower Ketone levels. I did not do any hard workouts during the testing window, and walked between 10,000 and 15,000 most days.   

 Ketogenic / LCHF Diet

By switching my preferred food source away from carbohydrates (specifically by eliminating highly processed carbohydrates yet embracing non-starchy vegetables) I am seeking to embrace much of the food of our ancestors, which is surprisingly high in fat and includes moderate amounts of protein. This has, over the course of a few weeks, allowed my body to ‘switch’ over to ‘burning’ fat. Once in ‘fat burning mode’ it is very happy to consume the many excess kilograms of visceral (and other) fat that I was carrying around. This has resulted in 20Kg of weight loss without deliberately restricting calories (except when fasting). I frequently measure my Ketone and Blood Glucose levels to help me know if I am on track. I did not realise that Tomato Sauce included a very high proportion of sugar, but the Ketone test executed after the meal with significant amount of tomato sauce focused my attention on what I had just consumed, and highlighted the problem. I must say it was quite uncomfortable getting into Ketosis, but once there, it is very easy to go without food. This gap in between eating may be a key component of the significant level of weight loss. Not needing to snack keeps my blood sugar low, which keeps in Insulin response subdued, which allows my body to consume fat for energy, rather than frequently having a high insulin level which signals my body to store the energy as fat for another day.

Results:

Triglycerides

The graph below shows what I am currently most interested in. It shows my Triglyceride level over number of hours since taking food. I understand that high levels of triglycerides effectively act on LDL particles, reducing them in size over time, until they become very small. These very small dense LDL particles appear to hang around in the blood, as the liver does not ‘want’ them, and their long term persistence exposes them to the possibility of oxidation or glycation, which increase the hazard that they represent to health. I am not particularly concerned about my LDL level, in and of itself, but am mindful that small dense LDL represents a serious health risk. I can’t directly measure these particles, but I can measure Triglycerides. If triglycerides are low, then there is less scope for the formation of small dense LDL. The graph shows that when my Ketone level is high, my Triglyceride level returns to it’s current base level faster than it does when my Ketone level is lower. The difference is in terms of Hours. If I have three meals in a day when in mild nutritional Ketosis, then the high triglyceride level could be high for twice as long as it would be with two meals per day, and a slightly deeper level of ketosis.

 LDL Cholesterol

This graph shows my (calculated) LDL level by hours since eating, with bubble size indicating the level of Ketosis. It shows that (for me) that my LDL numbers were better (lower) when Ketones were lower. This is why I’m not really that focused on the LDL level, but most doctors are. You can see from this graph that the 8 – 15 hour data points have a large range, so if I had a blood test in such a window, it could yield wildly different results – which could have a significant impact on treatment options. Yikes!!

Blood Glucose

This graph shows my blood glucose in the same way as the above graphs. I did notice during testing that early morning appear to have been impacted by a natural ‘cortisol effect’, where we get a boost to help us jump into our new day. This boost however signals the body to produce some glucose, which impacts both the Glucose and Ketone numbers. When waking after fasting for more than 40 hours, my blood glucose was very low (2.6mmol/L) so before eating, I went for a brisk walk, which lifted the blood sugar significantly. You can see this effect at the right side of this graph. It is really amazing how our body works.

Why It Matters (for me)

When triglycerides remain high, LDL particles are more likely to become very small and dense LDL which persist in the blood for longer times and can become oxidised and glycated and represent a real danger to health. These results help me realise that I am probably better off with 2 rather than 3 meals most days. I hope to see the fasting Triglyceride level gradually reduce of the coming months, and will be interested to see if the time taken for the level drops after eating changes, as my health improves. Such insights would not have been possible without running a relatively large number of tests in a short window of time, under a number slightly different situations. 

Please note that none of this information should be taken as advice, but is shared here for general interest, especially in the context of applying the testing based approach, (that is effective in building an understanding of complex IT systems) to ones own body, as a means of better understanding how it works and how to look after it.